Primary Cloud SQL instance is in Region1 with read replicas in Region2 and Region3. After promoting the read replica in Region2 during a Region1 outage, you want the same database capacity available before switching application connections. What should you do?

When a read replica is promoted to primary, it becomes a standalone instance. To restore the original database capacity (meaning two read replicas), you need to create new read replicas from this newly promoted primary. Creating one in Region3 re-establishes a replica in that region, and creating another in a different region (e.g., a new Region4 or even back in Region1 if it's recovered) provides the second replica, matching the original architecture. Enabling zonal high availability on the original primary (which is down) or creating a cascading replica from the remaining replica in Region3 doesn't restore the full read replica capacity from the new primary. Creating a new replica in Region1 and promoting it would be an unnecessary extra step and doesn't address the need for two read replicas from the current primary.